Simple Preamp (THAT / SSM / INA) - Development Diary

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

juliusbusch

Member
Joined
Jan 16, 2019
Messages
20
Location
Darmstadt - Germany
I'm currently developing a "new" simple preamp using the THAT 1510 IC and wanted to share my progress.

I want it to build myself a eight channel preamp into a 1U 19" rack format. The eight channel preamp will get eight TRS line outputs and a DB25 connector.
Each channel will have a combo jack with both, a XLR and a TRS input. The XLR input will have switchable phantom power for condenser microphones. There will be a gain potentiometer and a trimmer with  led metering before and after trimming.

2019-09-24-18-37.jpg


I'm not sure how to drive the output. Maybe I could use the THAT 1606 IC but this will get more expensive than using the NE5532. As I'd like to keep the cost as low as possible, what do you think so far?
 
If lowest cost is a priority just get rid of a line driver altogether.  An impedance balanced output is cheap and works well for most applications.
 
john12ax7 said:
If lowest cost is a priority just get rid of a line driver altogether.  An impedance balanced output is cheap and works well for most applications.
That is true, however there's an issue that needs to be resolved. Unless a suitable custom-made gain pot is ordered, those that are commonly available offer very poor resolution at higher gain, so the addition of a Trim stage is almost mandatory, but I agree, a simple impedance-balanced 5532 or similar is a perfect solution.
 
Group member Volker sold/sells some pcb's for dual mic amps that are not massively disimilar to what your designing. They are servo, with a balanced out all switching on relays, stepped gain switch.  I built  8 channels into a 1u with DB25 out they are very quite and very good !
 
john12ax7 said:
If lowest cost is a priority just get rid of a line driver altogether. An impedance balanced output is cheap and works well for most applications.
I've heard of the impedance balanced output. Thinking about it, are there any disadvantages regarding this method?

Rob Flinn said:
Group member Volker sold/sells some pcb's for dual mic amps that are not massively disimilar to what your designing. They are servo, with a balanced out all switching on relays, stepped gain switch. I built 8 channels into a 1u with DB25 out they are very quite and very good !
I'm sure that there are several members in this forum how have built their own preamps, but as I want to develop my own version, I won't buy any of those.
Do you have a link to his circuit? Maybe I can learn something, since I'm almost completely self thought concerning electronics.

Yesterday I did this PCB design.

2019-09-25-11-06%20top.png


2019-09-25-11-06%20bottom.png


Any thoughts on that?
 

Attachments

  • 2019-09-25-11-06 Simple Mic Preamp PCB Design.pdf
    48.2 KB · Views: 21
juliusbusch said:
I've heard of the impedance balanced output. Thinking about it, are there any disadvantages regarding this method?
No. Not any.

Yesterday I did this PCB design.
Why those bridgeable pads on the input connectors? Is it provision for input xfmrs? If so, I would rather put a 0.1" double-row HE10 connector.

I see you have connections from Pin 1 and shell to ground. This is wrong! These should go the chassis ground, via a sturdy connection. Some XLR sockets have a sharp pin that goes into the metal. That's the best chassis ground connection.

Why adding a second PCB for output and meter? It's more cost-effective to have a single board.
 
juliusbusch said:
I've heard of the impedance balanced output. Thinking about it, are there any disadvantages regarding this method?
Compared with a regular electronic balanced output:

1. Maximum output level will be 6dB lower (noise will be 3dB lower)
2. To maintain CMRR you need to reasonably accurately mirror the actual output impedance of the driver. In most cases this just means using the same value/type of output capacitor and series resistor.
3. You might also need to double the value of capacitors to maintain the same low frequency response.

Other than that its simplicity has a lot to recommend it.

Cheers

Ian
 
abbey road d enfer said:
Why those bridgeable pads on the input connectors? Is it provision for input xfmrs? If so, I would rather put a 0.1" double-row HE10 connector.

These contact i want to use as test points. They are labeled like their contact. I will close all of them, but leave the shield ("G") open. Of course, you are right about the shielding's purpose and I will use the Neutrik combo jack with metal spike.

abbey road d enfer said:
Why adding a second PCB for output and meter? It's more cost-effective to have a single board.

Good point! I will change this. I might do one board for eight inputs, with outputs and db-25 all in one.

ruffrecords said:
Compared with a regular electronic balanced output:

1. Maximum output level will be 6dB lower (noise will be 3dB lower)
2. To maintain CMRR you need to reasonably accurately mirror the actual output impedance of the driver. In most cases this just means using the same value/type of output capacitor and series resistor.
3. You might also need to double the value of capacitors to maintain the same low frequency response.

Other than that its simplicity has a lot to recommend it.

Cheers

Ian

I think, I will go with the differential line output, using the NE5532.
Thanks for all the information.
 
juliusbusch said:
I think, I will go with the differential line output, using the NE5532.
The issue there is that a discrete cross-coupled output stage is very difficult to optimize, even with precision resistors, so in practice they are less efficient than a well implemented impedance-balanced stage.
IMO, if you want a cross-coupled output stage, use one of the dedicated chips, such as THAT 1646.
 
juliusbusch said:
.....Any thoughts on that?
The parts on the schematic and pcb do not have same names. It's very hard to analyze it that way.

A capacitor at input RF filter C15, an electrolytic capacitor on phantom power (if there is any) and the ground pin on phantom power connector should be referenced to "chassis ground" (connected to pin 1) if the pin 1 is connected only to the chassis.
Footprints of capacitors C7 and 8 are too small, they should be 63V rated.
Why Rg and Cg are splitted, do you know what can be appropriate value and physical size of two Cgs for high gain?


 
So, I started all over again and  reconsidered some choices.
I now use the THAT 1646 output driver (thanks @abbey road d enfer) and hopefully got the whole audio / chassis ground thing right.

2019-10-08-00-18%20schaltung.jpg

2019-10-08-00-18%20pcb%20top.png

2019-10-08-00-18%20pcb%20bottom.png

X1 is the power connector for a bus cable.
X2 connects the LED bar in the front panel to the preamp PCB.
U1 is XLR input.
U4 is XLR output.
C9 is the CG (3300uF/10V) for preamp gain. (thanks @moamps)
TP5 is chassis ground.
Y1 and Y2 are wire bridges / jumpers for lifting ground references.
R5 is for bypassing the trimmer potentiometer.

All meters, potentiometers and switches will be mounted into the front panel and will be connected to the PCB using these 3-pin molex jacks.

The RF filters are mounted on the bottom side of the PCB, directly under the IO jacks.

---

What do you think?
 
juliusbusch said:
So, I started all over again and  reconsidered some choices.
I now use the THAT 1646 output driver (thanks @abbey road d enfer) and hopefully got the whole audio / chassis ground thing right.

R5 is for bypassing the trimmer potentiometer.
Read the datasheet for THAT1646; it must be driven with a very low impedance. You can't have a potentiometer there, or even a low value resistor.
 
You also should have a proper ground plane. And beware of putting an LED driver in close proximity to a mic pre. At least give it it's own ground. Otherwise, the rapid switching of LEDs dumping current into a common ground could be picked up and amplified by the v. high gain mic pre.

And just to add to abbey, you should run the 1510 directly into the 1646. The trim is just redundant since the mic gain control controls the level. Even if you want to have the option to insert another circuit, you could just require that it's output be low impedance. But if you want this to be "simple" you should skip the insert.

Note that, IMO, if there were one feature that almost every mic pre should have, I would say low-cut. You should consider that when you remove low frequencies, the dynamic range of your circuit is effectively improved greatly. Imagine a waveform where high frequencies are "riding" on top of the slow rolling low frequencies. If you remove the LF, your signal is much less likely to clip even at the same gain setting. This is why it's important to low-cut early in the mic pre.

So I would skip the insert and drop one of the LED drivers.
 
+1 for having a better ground design. You should use a larger track size for ground. "Proper" ground plane must take into account current circulation. You should not allow current from the LED's to pollute the preamp ground. That's why there is sometimes need to create "islands" in the ground plane to avoid "ground running in circles". If you're not familiar with this technique, I think it's simpler to use basic hierarchical ground (ground follows signal) with larger tracks.
 
abbey road d enfer said:
Read the datasheet for THAT1646; it must be driven with a very low impedance. You can't have a potentiometer there, or even a low value resistor.
Ok, trimmer and one VU meter are gone.

squarewave said:
Note that, IMO, if there were one feature that almost every mic pre should have, I would say low-cut. You should consider that when you remove low frequencies, the dynamic range of your circuit is effectively improved greatly. Imagine a waveform where high frequencies are "riding" on top of the slow rolling low frequencies. If you remove the LF, your signal is much less likely to clip even at the same gain setting. This is why it's important to low-cut early in the mic pre.
I will do that.

abbey road d enfer said:
+1 for having a better ground design. You should use a larger track size for ground. "Proper" ground plane must take into account current circulation. You should not allow current from the LED's to pollute the preamp ground. That's why there is sometimes need to create "islands" in the ground plane to avoid "ground running in circles". If you're not familiar with this technique, I think it's simpler to use basic hierarchical ground (ground follows signal) with larger tracks.
Sadly I'm not familiar with those "special" grounding methods. I will do a ground plane for the preamp and a separate one for the meter(?). What do you think about not placing the meter circuit on the same PCB at all?
 
juliusbusch said:
I will do that.
Note that, if you add a HPF, it can be driven by a potentiometer, provided it's value is low enough compared with the resistors in the filter. Typically, a Sallen & Key HPF with 50k and 25k resistors works perfectly with a 10k pot. Difference in slope and frequency between max gain and -6dB is negligible.

Sadly I'm not familiar with those "special" grounding methods. I will do a ground plane for the preamp and a separate one for the meter(?).
That would be a good start.

What do you think about not placing the meter circuit on the same PCB at all?
You can create two (or more) polygons with Eagle; give them two different node names and connect them to ground via zero-ohm links.
 

Attachments

  • HPF at max and -6dB.jpg
    HPF at max and -6dB.jpg
    72.8 KB · Views: 38
abbey road d enfer said:
Note that, if you add a HPF, it can be driven by a potentiometer, provided it's value is low enough compared with the resistors in the filter. Typically, a Sallen & Key HPF with 50k and 25k resistors works perfectly with a 10k pot. Difference in slope and frequency between max gain and -6dB is negligible.
Where schould I position this "Sallen & Key HPF"? Behind the 1510/12?
Can I use a passive RC Filter at the two input phases instead? This would be less components.
 
juliusbusch said:
Where schould I position this "Sallen & Key HPF"? Behind the 1510/12?
Yes. Pot first, then filter.

Can I use a passive RC Filter at the two input phases instead?
I wouldn't recommend it. Adds noise and reacts differently on different mics.

This would be less components.
Ah! the attraction of minimalism... If it takes ten components to do the job properly, and three to do it mediocrely, I'd choose the former.
 
2019-10-09-19-41%20pcb.jpg

2019-10-09-19-41%20schaltung.jpg

2019-10-09-19-41%20pcb%20top.png

2019-10-09-19-41%20pcb%20bottom.png

R18 now is the Link between the VU meter ground and the audio ground
and the HPF is between 1510 and 1646.
I only did ground planes in the top layer. Will I have to ad them in the bottom layer, too?
(And yes, I forgot connecting the filter's opamp to supply voltage, in these pictures)
 
juliusbusch said:
I only did ground planes in the top layer. Will I have to ad them in the bottom layer, too?
You should increase the width of the ground tracks; it makes routing somewhat more difficult, but that's what double-sided is for.
I noticed Pin 1 of the output connector is connected to audio ground; should be connected to "earth".
Have you breadboarded the LM3914 circuit? I'm not sure you'll find it satisfactory...
 
Back
Top